这几天观察:Google、OpenAI、Meta、xAI 都推了个人 AI agent,Anthropic 好像还没单独做一个「全能助手」。
有人说 Claude 落后了;也有人说 Claude Code / Projects 已经够用,不必再堆一个聊天机器人。
你站哪边?日常更想要「一个什么都能干的 agent」,还是「只在写代码/项目里特别强」?
@sama Custom UI answers are exciting when the task is visual or interactive. For dense analysis I still want plain text first—curious how often people keep the generated UI vs copy the text out.
@bscholl This might be the most underrated framing of AI economics: the new divide won't be capital vs labor, but people who compound effort against those who don't. 效率红利越大的时候,这个分化越明显。
@Variety Fair point: performance rights should be an explicit, compensated license, not a boilerplate AI waiver buried in the contract. 想看后续片方会不会把 AI 条款单列出来谈。